CIU Logo
 
Interim 2021-based Population Projection Results
Demographic Update
January 2023
 
 
 
 
Executive Summary

This summary document outlines the results of the GLA’s 2021-based population projections which were released in January 2023.

These projections take the 2021 mid-year estimate (MYE) population as their starting point. The 2021 MYE, published by ONS in December 2022, is the first to incorporate the results of the 2021 census. As the 2021 MYE does not align with the current official estimates for 2012-2020, the GLA has based these projections on an alternative modelled series of population and components that is consistent with the 2011 and 2021 estimates.

This release is considered interim as some of the key data required to fully update the projections are not yet available. We anticipate the projections will be updated later in 2023 following revisions by ONS to the 2021 mid-year estimates and the release of additional data from the 2021 census.

The Interim 2021-based population projections comprise both variant trend projections and housing-led scenario projections. Trend projections are based only on the extrapolation of past patterns of population change while the housing-led projections also incorporate considerations of future housing delivery. Trend projections are typically most robust at larger geographic scales, such as regions or subregions. For applications that require finer spatial detail, the GLA recommends the use of the housing-led projections.

Trend projections

The 2021-based trend projections include three variants which differ in their assumptions about future migration behaviour (both international and domestic). These assumptions are based on averaging past periods of observed migration behaviours. The variants are:

Trend projections are available for all local authority districts in England and Wales and at the national level for Northern Ireland and Scotland.

Housing-led projections

The 2021-based housing-led projections include three scenario projections based on different rates of future housing development. They are:

In producing the projections no assumption has been made about the likelihood of these delivery scenarios coming forward - the projections are designed to illustrate how the population might change under the conditions described in the scenarios. At the London level all housing-led projections are constrained to match the 10-year trend projection and differ only in terms of the spatial distribution of population.

Due to limited information about potential development sites in the very long-term, the housing-led projections have a projection horizon of 2041.

Housing-led projections are available for London Boroughs and London wards (2022 boundaries).

Key results
Why the projections show lower levels of future growth

Recent rounds of GLA projections have shown progressively lower rates of future growth in London. This is a consequence of lower levels of annual growth in the population estimates on which the projections are based. Relative to the exceptional period from 2004 to 2016, when the population of London grew by more than 100 thousand per year, estimates for years since 2016 have shown far lower rates of growth. As this period of lower growth has continued, its contribution to the trends of future patterns projected forward in our modelling has increased, leading to lower projections of future growth.

The impact of recent lower growth is reflected most strongly in projections based on short periods of past data. This is clearly illustrated by the very low growth projected in the 5-year trend variant of the current release, for which future migration patterns are based entirely on the period of lower growth that began in 2016.

An explanation of why recent estimates of growth have been falling in recent years is perhaps most easily broken down into three sections:

Factors resulting in lower growth from 2016

Prior to the start of the COVID-19 pandemic, slowing rates of population growth were the result of several trends:

Impacts of the COVID-19 pandemic

The COVID-19 pandemic had a dramatic impact on both mortality and migration patterns, leading to the first annual fall in London’s population since the 1980s (see Population change during the pandemic).

While there is considerable uncertainty about precisely how much the population fell as a result of the pandemic, the modelled backseries produced by the GLA to underpin the projections assumes a decrease of approximately 80 thousand over the year to mid-2021.

While a fall of this scale (108 thousand) was anticipated in the GLA’s 2020-based projections, it did not factor into the past trends used to determine long-term migration assumptions. For the 2021-based projections, data for the year to mid-2021 directly contribute to projected patterns of future migration, resulting in a relative reduction in long-term growth rates.

Effect of re-basing population estimates following the 2021 Census

The 2021 mid-year estimates published by ONS in December 2022 were the first to incorporate data from the 2021 census, with the population for mid-2021 rolled-forward from the census day estimate rather than from existing 2020 estimate (see London’s Population and our Census Estimates Blog).

The 2021 MYE gave London’s total population as 8.80 million, 206 thousand less than in the 2020 MYE. This difference represents the combined impact of several factors:

While there is significant uncertainty about how the difference between the 2020 and 2021 MYE populations should be distributed between these factors, the impact of incorporating the 2021 estimate into the modelling is to reduce projected growth relative to projections based only on the original estimates series.

1 Introduction

Back to top

The GLA Demography team produce a range of population projections which are used to help inform planning and policy work within the GLA and across London. Projections are updated annually, incorporating the latest data as it becomes available.

This summary document outlines the results of the GLA’s Interim 2021-based population projections which were released in January 2023.

The 2021-based projections take as their starting point the most recent ONS Mid-Year Estimate (mid-2021). This is a rolled-forward estimate from the 2021 census estimate. A backseries of population and components of population change for the period 2012-2020 has been modelled by the GLA for use in these projections.

The two main types of population projection produced are trend-based projections, which are based on a cohort component model that uses past trends in births, deaths and migration to project future populations, and housing-led projections, which reconcile future population growth with forecast housing supply.

The trend projections comprise a set of variants based on different assumptions about future migration levels. The housing-led projections include a set of scenarios across which the migration assumptions are consistent but the levels of housing delivery in London differ. The production of these variants and scenarios has been driven by the desire to meet the needs of a wide range of users.

In producing the projections no assumptions have been made about the likelihood of these migration variants or housing delivery scenarios coming forward - the projections are designed to illustrate how the population might change under the conditions described in the variants/scenarios.

Detailed methodology documents are provided alongside each set of population projections which can help users decide which set of population projections is most appropriate for their needs.

2 Trend projections

Back to top

Three variant projections have been produced: a central projection, and two projections which bound a wider plausible range. These variants have been produced in order to assist users in understanding current uncertainty about future population growth. The projections differ in their assumptions about future levels of domestic and international migration, they are:

The projections move forward from a starting point of mid-2021. In the first year of the projection (2022) international migration levels are constrained to be consistent with UK-level data on international flows published by ONS.

2.1 Expert Panel

Back to top

For recent rounds of population projections the GLA sought the guidance of an expert panel of academics and professional demographers in setting the projection input parameters. The panel were consulted on the economic and social drivers of population change in both the short-term and long-term, as well as the specifics of component trajectories. Their input was key in shaping the assumptions which formed the 2019-based and 2020-based variant projections. A summary of the consultation responses and recommendations is available on the London Datastore.

For the 2021-based projections the GLA has returned to a statistical approach to assumption-setting. We believe that the three trajectories, based on varying periods of past migration data, provide a reasonable range of likely future population change. As such, the expert panel’s input was not sought these interim projections, though we anticipate seeking input from external experts once again for the 2022-based projections.

3 Interim projections

These 2021-based projections, released in January 2023, are interim projections while we await the publication of additional 2021 census datasets, updates to the 2021 Mid-year Estimate (MYE) and the official ONS MYE backseries for the period 2012-2020.

4 Population backseries

The production of population projections requires an annual backseries of population and components of change. The most recent mid-year estimate (MYE) is used as the ‘jump-off’ point for the projections and data from past years’ estimates is used in the calculation of average flows and rates.

The 2020-based projections used the 2020 MYE as their jumping-off point. This was the last estimate in a series which began in 2011. Each subsequent year from 2012 forward was calculated by adding estimated components of change to the previous year’s population estimate. As such, over the decade, the MYE series had drifted away from the true population count.

The 2021 census has provided an opportunity to reset the mid-year estimate series using the new data census as a starting point. The 2021 estimate therefore represents a break in the previous series. In order to create a consistent back series between the 2021 and 2011 estimates the GLA has modelled population and components of change for the period 2012-2020.

The existing ONS 2020 MYE for London was 9.00 million. The GLA’s revised estimate for 2020 is 8.88 million, around 122 thousand persons lower. This difference represents the combined impact of several factors:

Regardless of how the differences are distributed between these factors, the impact on the projections is significant. The 2021-based projection, starting from a lower population and based on lower growth trends over the last decade, have lower future growth trajectories than the 2020-based projections.

5 Interpreting variant projections

Back to top

The 2021-based variant projections provides users with three population trajectories. These variants give an indication of the effect of different levels of migration on population size and structure.

The last two decades have seen a number of shocks which might be expected to have significantly impacted international flows: EU expansion in 2004, 2007 and 2013; the credit crunch, housing crisis and great recession beginning in 2008; the EU referendum in 2016; and major changes to UK migration policy throughout the period. However, absolute levels of international migration have been relatively stable, lending weight to the idea of the resilience of long-term trends.

The three 2021-based projection variants use differing levels of migration - both domestic and international - and assumes that that level remains consistent throughout the projection period. Projected domestic flows are calculated by applying averaged rates of migration while international in and out migration are input as static flows.

The 10-year trend average captures a wide range of behaviours across the economic cycle, including the lower levels of mobility observed following the financial crisis. The five-year average provides a higher migration scenario - the last five years have been a period of relatively high migration with a peak in 2017 when net out migration from London reached 107 thousand people. Use of the five-year average represents an assumption that recent levels become the norm even after accounting for potential variation over the economic cycle.

The inclusion of data covering the period from mid-2019 to mid-2021 in the averaging was carefully considered. Although the data for these years does include the impacts of COVID-19 on population movement, the range of variants provides users with a clear indication of the range of plausible future growth when migration inputs are varied.

The 10-year trend projection projects annualised growth of 42 thousand per year over the period to 2050. This is in the context of average growth of 59 thousand per year in London over the last decade.

Table 1: Annualised population change, London

5.1 Projections for areas outside London

The detailed variant projections outputs include data for all local authorities in England and Wales as well as national-level data for Northern Ireland and Scotland. This report presents the results for London only. Data for areas beyond the Greater London boundary are primarily published for the purpose of transparency and do not necessarily reflect the organisation’s view of future growth outside of the city.

6 Methodology

Back to top

A paper on the models, methodologies and configuration options used in producing these projections is available to download from the London Datastore.

7 Trend projections results

Back to top

7.1 Projections in the Short-term

The initial short-term period of the projections incorporates additional intelligence about population change not included in the mid-year estimate series. In the first projection year (2022) international migration at the UK level is adjusted to be consistent with data from the ONS on flows for that year. This year saw extremely high in-migration flows as a result of significant flows from Ukraine and Hong Kong and a ‘bounce back’ following the lifting of covid restrictions.

Table 2: International migration assumptions, 2022

Over the following two years (2023-2024) migration assumptions diverge as levels of migration transition incrementally towards the long-term values set for each variant. These long-term levels come into effect in 2025.

7.2 Long-term projections

Beyond 2025, long-term levels of migration are reached and maintained.

In the 10-year trend projection, international migration is net 78 thousand per year while in the 5-year trend variant international migration is 64 thousand per year and in the 15-year trend variant 83 thousand per year.

All three variants project population growth over the period, albeit to varying levels. London grew by 1 million people in in the decade to 2011 and then around 800 thousand people in the the decade to 2021. Over the decade to 2031, the 10-year trend Population variant sees growth of 602 thousand while the 5-year trend and 15-year trend grow by 289 thousand and 735 thousand respectively.

The 5-year trend represents a somewhat extreme scenario taking as it does only very recent migration trend which include the joint shock of COVID-19 and Brexit. It the other scenarios these low net migration years are counterbalanced by the higher levels of migration included in the longer averages. The 5-year trend is potentially useful for short-term planning however the GLA recommends longer-term projections for longer-term strategic planning.

Table 3: Variant population projections, London
Source: GLA 2021-based population projections

8 Comparison to 2020-based projections

Back to top

The 2020-based GLA trend projections were released in September 2021. The differences between the last round and the 2021-based projections are significant, primarily because the population backseries used differs.

The 2021-based projections use the 2021 MYE as their starting point and the modelled GLA backseries for the calculation of trends. The 2020-based projections used trends based on the ONS series which had overestimated the level of population growth in London. As a result, the previous set of projections started from a higher population and projected stronger growth into the future when compared to the 2021-based projections.

8.1 Population

In the 2020-based projections the Central Range of population in 2050 was 10.319 million to 10.707 million, while for 2021-based projections growth is projected to range between 9.24 million (5-year trend) and 10.38 million (15-year trend). The 10-year trend variant projects a population of 10.06 million in 2050.

8.2 International migration

The 2020-based projections maintained a long-term net international migration level of 95 thousand for both the Central Upper and Central Lower projections. In the 2021-based projection each variant has a lower level of net international migration in the long-term (between 64 thousand and 83 thousand). The difference comes from the different levels of past migration used to calculate the flows. The 2020-based projection used higher past migration assumptions when calculating average flows (see chart) and so projects higher future flows.

8.3 Domestic migration

The difference between the 2020-based and the 2021-based projections is less pronounced in the domestic migration component. Both projections include a ‘bounce-back’ in net domestic in the year to 2022. In the 2020-based projection net domestic goes from -150 thousand in 2021 to -45 thousand in 2022, while in the 2021-based projections it goes from -179 thousand to -93 thousand (10-year trend).

8.4 Births

The GLA projection model incorporates fertility assumptions from the ONS National Population Projection. The annual change in age-specific fertility rates is applied rates at the local authority level to create local authority-specific trajectories for the projection period. The 2020-based trend projections used the 2018-based NPP principal trajectory in this process. The 2021-based projections use the 2020-based principal trajectory. This change accounts for much the difference in births, particularity in the initial years of the projections.

8.5 Deaths

As with fertility rates, a set of age-specific mortality rates are calculated for the projection period using trajectories from the NPP. The 2020-based projections used the principal trajectory from the 2018-based NPP while the 2021-based projections incorporate data from the 2020-based NPP.

8.6 Natural change

Natural change is the largest contributor to London’s population growth. In the past it has ranged as high as 87 thousand in 2011. Natural change is higher in the 2021-based 10-year trend and 15-year trend variants because of the higher level of births and lower level of deaths in these projections.

9 Age structure

The age structure in 2037, after 15 years of projected change, is shown in the chart below.

10 Working age population

The size of the working age population (16-64) in London is heavily dependent on the level of migration. High net out migration as seen in the 5-year trend variant leads to a decline in the working age population over time. In the 15-year trend variant, which has a low net out migration, the population grows over the period to 2050.

The 10-year trend variant sees growth up to 2040 and then a leveling-off in the last decade of the projection.

Table 4: Working age population, London

11 Components of Change

Back to top

Births, deaths and migration all contribute to London’s changing population. Natural change, which is the difference between the number of births and deaths, is the largest direct contributor to London’s population growth. Natural change is comparatively high in London because its age structure is much younger than that of the rest of the UK meaning more births and fewer deaths.

As noted above, London’s relatively youthful population is a result of established patterns of migration to and from the capital, which give rise to a net inflow of young adults and a net outflow of all other age groups.

11.1 Total migration

Total net migration is the sum of all flows in to and out of an area. This includes both international and domestic flows. Historically, London has seen large gross flows both in and out of the capital which have largely cancelled each other out. Over the period 2002-2019 total net migration ranged between -38 thousand thousand and 63 thousand.

COVID-19 saw significant disruption to migration flows both domestically and internationally. In particular, 2021 saw significantly higher than normal net outflows from London (-135 thousand). Flows in the initial year of the projection (2022) include data from ONS on international migration at the UK level. The year to mid-2022 saw the highest net inflow of international migrants to the UK on record (504 thousand). This was in part a response to the easing of covid restrictions but was also influenced by significant migrations from Ukraine and Hong Kong. This UK-level data has been used to model a net international inflow to London of 170 thousand in 2022.

11.2 International migration

Net international migration in the 15-year trend scenario is 83 thousand persons per year, in the 10-year trend scenario is 78 thousand persons per year and in the low scenario is 64 thousand persons per year. These levels are achieved in 2025 in the modelling. The period 2020-2022 has its own international migration profile which is consistent across all three scenarios. The period 2023-2024 is a transition period over which levels of migration move from the 2022 level to the 2025 level.

11.3 Domestic migration

The balance of domestic migration in London has been consistently negative for the last two decades. This means that more people leave the capital each year for other areas of the UK than move to London. Higher rates of domestic migration overall lead to a higher net outflow from London and correspondingly slower population growth.

Over the last decade average outflow from London was 93 thousand. The average of the last 15 years is lower at 79 thousand as this period includes the Great Recession when domestic flows were lower across the UK in response to economic uncertainty and impacts on the housing market. In more recent years the combined impacts of COVID-19 and Brexit have meant significantly higher levels of domestic out migration from London. The 5-year average outflow for the period 2017-2021 was 117 thousand.

11.4 Births

Annual births in London rose by approximately 30 per cent over the decade from 2002 to 2012. Since then the number of births has fallen and in 2021 was 110 thousand, similar to the level last seen in 2003.

Total births in 2022 are modelled from GP register counts of 0-year-olds and are estimated to be slightly higher than the number recorded in 2021 at 111 thousand. Births in subsequent years are projected by applying projected fertility rates to projected populations. Fertility rates in 2023 are calculated by trending forward rates over the 5-year period 2018-2022. In subsequent years that initial rate is adjusted by the rate of change taken from the ONS 2020-based NPP principal projection assumptions.

The same fertility rates are applied in each of the variants. The difference in births is a function of the differing size and composition of the population in each variant. In all cases an initial period of falling births is followed by steady increase.

11.5 Deaths

The same mortality rates are applied in each of the variants. The difference in deaths is a function of the differing size and composition of the population in each variant. Mortality rates in 2022 are calculated by continuing the 5-year trend over the period 2015-2019 (i.e. pre-COVID-19). Rates for 2023 and subsequent years are calculated by adding the rate of change from the ONS 2020-based NPP to the borough-specific rates.

The impact of COVID-19 can be seen in the deaths occurring over the period 2019-2021. From 2022 on deaths from COVID-19 are assumed to be accounted for through the normal application of mortality rates.

11.6 Natural change

Natural change is the balance of births and deaths over the year. It is the largest direct contributor to London’s growth. In 2021 natural change added 55 thousand persons to the London population.

12 Housing-led projections

Back to top

The housing-led projections are produced for London boroughs and wards. The GLA suggests that these projections provide the best estimate of future populations at these geographic resolutions. This is because the additional intelligence on the capacity of smaller areas to accommodate population, in the form of available housing stock, provides a more realistic view of future population. The GLA recommends the housing-led projections for most purposes.

Unlike the trend projections which have a projection horizon of 2050, the housing-led projections work to a projection horizon of 2041. This is because the availability, reliability and utility of housing delivery data diminishes over time.

The 2021-based housing-led projections comprise three scenario projections each of which uses a different housing delivery trajectory. In producing the projections no assumption has been made about the likelihood of these delivery scenarios coming forward - the projections are designed to illustrate how the population might change under the conditions described in the scenarios.

12.1 Identified Capacity Scenario

Back to top

This scenario adapts the 2017 Strategic Housing Land Availability Assessment (SHLAA) housing capacity study for use as a future delivery scenario. A SHLAA is a technical exercise to determine the quantity and suitability of land potentially available for housing development. It is a required part of the evidence base needed for the preparation of a Local Plan.

In the initial years of the projection the delivery assumptions have been amended to reflect housing market analysis undertaken by Savills Estate Agents. This is in lieu of completions data for the years 2020-2025 which we are not yet able to incorporate into the projections. In 2020 assumed delivery is 41,700 units while for 2021-2025 it is 43,000 per annum. The total delivery over this period is 256,700 while the SHLAA capacity identifies 333,100 units over the same period. The difference (76,400) is assumed to come forward over the period 2026-2041 and is distributed to the trajectory for those years.

In later years the trajectory assumes lower levels of delivery. Whilst this is unlikely to be a reality for London’s future growth, it is a reflection that the SHLAA, on which this delivery scenario is based, is a capacity study that - due to information available on site availability and deliverability at the time - is better able to identify sites in the short-to-medium term. In the longer-term SHLAA capacity is based on less reliable assumptions about sites that may come forward for housing delivery. This is due to an inability to predict market changes or development patterns that could result in brownfield land not currently suitable for housing becoming available. As a result, annual delivery in later years is much lower than the initial years of the trajectory, however in reality this is unlikely to occur.

In the Identified Capacity scenario, the number of dwellings in London is forecast increase by 788 thousand over the 21-year period to 2041. This is an increase of 21.6% over current stocks.

12.2 Past Delivery Scenario

Back to top

This scenario assumes that the average of past levels of development, as reported in the London Development Database (LDD) will come forward each year for the period 2022-2041.

In the initial years of the projection, the level of delivery is assumed to be the same as in the Identified Capacity scenario: 41,700 in 2020 and 43,000 in 2021. The underlying assumption being that this recently developed intelligence, undertaken in 2021 during the pandemic, provides the most reasonable estimate of short-term delivery. Beyond 2021, the LDD average of 31,100 units is used.

In this scenario, the number of dwellings is forecast to increase by a total of 664 thousand units over the 21-year period to 2041. This is an increase of 18.2% over current stocks.

12.3 Housing Targets Scenario

Back to top

This scenario assumes that the London Plan target of 52,300 dwellings will be delivered in each year of the projection period up to 2041. In the initial years of the projection, the level of delivery is assumed to be the same as in the Identified Capacity scenario: 41,700 in 2020 and 43,000 in 2021. The underlying assumption being that this recently developed intelligence, undertaken in 2021 during the pandemic, provides the most reasonable estimate of short-term delivery.

In the Housing Targets scenario, the number of dwellings is forecast to increase by 1.089 million units over the 21-year period to 2041. This is an increase of 29.8% over current stocks.

12.4 Development Trajectories

12.4.1 Assumed development

The chart shows the net additional dwellings assumed in each year of the projection for the three scenarios. Data for 2012-2019 are taken from the LDD.

12.4.2 Assumed stock

The chart shows the cumulative growth in units over the projection period in each of the three scenarios.

13 Housing-led projections results

Back to top

13.1 Comparison to the Trend

The housing-led projections are constrained to be consistent with the 10-year trend Trend projection at the International Territorial Level 2 (ITL2, previously NUTS2) aggregation level. The ITL2 sub-regions are:
- Inner London - West
- Inner London - East
- Outer London - East and North East
- Outer London - South
- Outer London - West and North West

As a result of this constraining, the population of each group of boroughs, and of London as a whole, is the same in all of the housing-led projections. Individual borough and ward populations do vary.

13.2 Population change

Back to top

The housing-led projections are recommended for most uses, especially those at borough level. The maps below show the spatial distribution of growth over the projection period 2020-2041 in each of the three scenarios.

13.2.1 Identified Capacity

Map 1: Population growth 2020-2041, Identified Capacity Scenario
Source: GLA 2021-based population projections

13.2.2 Past Delivery

Map 2: Population growth 2020-2041, Past Delivery Average Scenario
Source: GLA 2021-based population projections

13.2.3 Housing Targets

Map 3: Population growth 2020-2041, Housing Targets Scenario
Source: GLA 2021-based population projections

13.3 Average Household Size

Back to top

Average Household Size (AHS) is calculated by dividing the household population of an area by the number of households. These levels of AHS are adjusted in the model to account for past trends in population change and the amount of available housing. The results can be seen in the chart below. The Past Delivery scenario has lower housing delivery meaning population must fit into fewer households and thus raising AHS compared to the other scenarios.

13.4 Ward projections

Back to top

Housing-led projections are also available for London wards. These projections use a modelled backseries based on ONS small area population estimates (SAPE) and consistent with the GLA revised MYE backseries. Data for 2021 may not be consistent with census estimates for wards. Future updates to projections will incorporate ward-level information from the census.

Summaries for each ward as well as detailed projections and estimates are available on the London Datastore.

 
 
 
For more information please contact the Demography Team, GLA City Intelligence Unit.
demography@london.gov.uk
 
Copyright © Greater London Authority, 2023